ncbo / ontologies_api_ruby_client Goto Github PK

Hi,

I don't know if you had already saw this issue.

The issue

At Agroportal, we are using the tag 2.0.2 (as you, i think) and recently when we did a bundle update we got the last version of faraday 2.1.0 where the multipart middleware (that is used here) is no more present (so causing a runtime error, see screenshot below ), it was moved to a separate gem called faraday-multipart.

More infos here : https://github.com/lostisland/faraday/blob/main/UPGRADING.md#bundled-middleware-moved-out

The fix for tag 2.0.2

For the tag 2.0.2 (used in production), we solved it by fixing the faraday version in the gemspec to 1.4.3

Gem::Specification.new do |gem|
  gem.authors       = ["Paul R Alexander"]
  gem.email         = ["[email protected]"]
  gem.description   = %q{Models and serializers for ontologies and related artifacts backed by 4store}
  gem.summary       = %q{This library can be used for interacting with a 4store instance that stores NCBO-based ontology information. Models in the library are based on Goo. Serializers support RDF serialization as Rack Middleware and automatic generation of hypermedia links.}
  gem.homepage      = "https://github.com/ncbo/ontologies_api_ruby_client"

  gem.files         = `git ls-files`.split($\)
  gem.executables   = gem.files.grep(%r{^bin/}).map{ |f| File.basename(f) }
  gem.test_files    = gem.files.grep(%r{^(test|spec|features)/})
  gem.name          = "ontologies_api_client"
  gem.require_paths = ["lib"]
  gem.version       = "2.0.2"

  gem.add_dependency('activesupport', '~> 5.2.6')
  gem.add_dependency('excon')
  gem.add_dependency('faraday', '1.4.3')
  gem.add_dependency('lz4-ruby')
  gem.add_dependency('multi_json')
  gem.add_dependency('oj')
  gem.add_dependency('spawnling', '2.1.5')

  # gem.executables = %w()
end

The fix for tag 2.0.3

The fix for the last tag is to add the faraday-multipart gem as a dependency in the gemspec.

Hi,
Within this repo, I have done three PRs where the tests worked in our forked environments but not here

Example

My latest PR with the same code

In ontoportal-lirmm

In NCBO

The purl method is generating URLs that result in 404 Not Found. Some examples:

osteochondrosis (from RADLEX)
PURL: http://purl.bioontology.org/ontology/RADLEX?conceptid=http%3A%2F%2Fradlex.org%2FRID%2FRID5369

Abnormal Cell (from NCIT)
PURL: http://purl.bioontology.org/ontology/NCIT?conceptid=http%3A%2F%2Fncicb.nci.nih.gov%2Fxml%2Fowl%2FEVS%2FThesaurus.owl%23C12913

The PURLs will resolve properly if the conceptid parameter's value isn't URL encoded.

I'm testing the BioPortal Rails application against Ruby 3.0.6, and the staging version of the BioPortal REST API, which is currently running on the AllegroGraph database. In my local dev environment, I'm seeing intermittent failures from two of the API calls inside of the refresh_cache method:

def refresh_cache
  Spawnling.new do
    LinkedData::Client::Models::Ontology.all
    LinkedData::Client::Models::OntologySubmission.all
    LinkedData::Client::Models::User.all
    exit
  end
end

The call to fetch all OntologySubmission objects results in an empty array, or a Faraday::TimeoutError. The call to fetch all users sometimes returns user data, and sometimes results in the same Faraday::TimeoutError.

The responses of these calls are never examined/logged, so they essentially appear to be silent failures. @mdorf is looking into why the /submissions endpoint returns an emtpy array from an AllegroGraph database (see ncbo/ontologies_api#117).

We may want to consider - at a minimum - logging error responses from these calls, so that we're at least aware when they occur.

We've had several reports from the developers of the EDAM ontology where they upload a new ontology submission and the user interface only shows the submission status as uploaded. After waiting (sometimes as long as 24 hours), they're never able to see the status as anything other than uploaded, i.e., the data for the new submission never becomes visible.

I've reproduced this behavior in our production environment on two separate occasions, and observed the following:

• It doesn't matter if you are logged in our out. The new submission shows as "uploaded" for both cases.
• Clearing the browser cache has no effect.
• Manually clearing BioPortal's goo and HTTP caches from the administrative page has no effect.

The only way I was able to get the new data to become visible in the UI was to click the "Flush UI Cache" button on the administrative page, which clears the front end in-memory cache (memcached).

I've also verified (by looking at the production parsing logs), that the EDAM ontology is relatively small and the system has no difficulty during the parsing process. From start to finish, we process the ontology in roughly 12 minutes.

Refactor existing unit test code to use the more modern minitest gem for testing.

Currently, we are using production Bioportal api for unit testing of ontologies_api_ruby_client which limits the kind of unit tests we can safely run. Instead of this we should spin up a dockerized and dedicated ontologies_api for unit testing.

It appears that the gemspec file in this project hasn't been actively maintained, i.e.:

• Since we're using this code in production, the version number should be (at least) 1.0.0.
• We haven't been bumping the version number with code changes.
• The email and summary fields should be updated with more current information.

I couldn't find any functionality in the BioPortal RoR application that triggers a call to find on LinkedData::Client::Models::Class, nor do we have any unit tests that exercise that particular method.

If you attempt to call the method in a debugging environment, e.g.:

LinkedData::Client::Models::Class.find(
  'http://purl.bioontology.org/ontology/STY/T047', 
  'https://data.bioontology.org/ontologies/STY'
)

... the library throws an ArgumentError:

ArgumentError: wrong number of arguments (given 1, expected 0)

The method attempts to explore a non-existent type of hypermedia link on an ontology object (i.e., 'class' instead of 'single_class').

We're currently using the Spawnling gem to fork a new process for refreshing the caches. The gem appears to have been abandoned in 2015. We should find / use a replacement, e.g., Sidekiq.

Received a complaint on the support list from a user that tried to remove an administrator from their ontology via the "Edit Ontology Information" form in the BioPortal UI, and received a 500 error. This form is accessed in the UI by going to any ontology summary page, e.g., http://bioportal.bioontology.org/ontologies/ECSO, and clicking the "Edit ontology information" link.

The production log file shows a Faraday::Timeout error:

I, [2017-04-05T10:47:07.646950 #21412]  INFO -- : Completed 500 Internal Server Error in 60274ms (ActiveRecord: 0.0ms)
F, [2017-04-05T10:47:07.649109 #21412] FATAL -- :
Faraday::TimeoutError (read timeout reached):
  app/controllers/ontologies_controller.rb:398:in `update'

I, [2017-04-05T10:47:07.650174 #21412]  INFO -- : Processing by ErrorsController#internal_server_error as HTML
I, [2017-04-05T10:47:07.650307 #21412]  INFO -- :   Parameters: {"utf8"=>"✓", "authenticity_token"=>"", "ontology"=>{"name"=>"The Ecosystem Ontology", "administeredBy"=>["", "http://data.bioontology.org/users/brycemecum", "http://data.bioontology.org/users/mobb", "http://data.bioontology.org/users/schild"], "viewingRestriction"=>"public", "acl"=>["", "http://data.bioontology.org/users/cjones", "http://data.bioontology.org/users/mobb", "http://data.bioontology.org/users/schild", "http://data.bioontology.org/users/sophisticus", "http://data.bioontology.org/users/xixiluo"], "hasDomain"=>["", "", "http://data.bioontology.org/categories/Other"], "isView"=>"0", "subscribe_notifications"=>"1"}, "commit"=>"Save ontology", "id"=>"ECSO"}
I, [2017-04-05T10:47:07.655224 #21412]  INFO -- :   Rendered errors/internal_server_error.html.erb within layouts/ontology (2.0ms)
I, [2017-04-05T10:47:07.658264 #21412]  INFO -- :   Rendered layouts/_topnav.html.haml (1.8ms)
I, [2017-04-05T10:47:07.658598 #21412]  INFO -- :   Rendered layouts/_notices.html.erb (0.2ms)
I, [2017-04-05T10:47:07.658686 #21412]  INFO -- :   Rendered layouts/_header.html.erb (3.2ms)
I, [2017-04-05T10:47:07.659514 #21412]  INFO -- :   Rendered layouts/_footer.html.erb (0.7ms)
I, [2017-04-05T10:47:07.659725 #21412]  INFO -- : Completed 500 Internal Server Error in 9ms (Views: 7.5ms | ActiveRecord: 0.0ms)

Just FYI - I removed the value of the authenticity token in the parameter list above.

Calling the update method in this project on an ontology object results in a read timeout. I wasn't able to reproduce this behavior after several attempts. We'll need to investigate under what circumstances the timeouts occur.

After upgrading to Ruby 2.7, the log files for the BioPortal RoR application are full of URI.escape is obsolete warnings, originating from this library, e.g.:

Started GET "/ajax/classes/label?ontology=NIFSTD&concept=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FUBERON_0000019" for 127.0.0.1 at 2022-12-19 11:34:52 -0800
Processing by ConceptsController#show_label as */*
  Parameters: {"ontology"=>"NIFSTD", "concept"=>"http://purl.obolibrary.org/obo/UBERON_0000019"}
/Users/vendetti/.rbenv/versions/2.7.7/lib/ruby/gems/2.7.0/bundler/gems/ontologies_api_ruby_client-ca18863efae6/lib/ontologies_api_client/link_explorer.rb:64: warning: URI.escape is obsolete
  Rendering text template
  Rendered text template (Duration: 0.0ms | Allocations: 6)
Completed 200 OK in 362ms (Views: 0.3ms | ActiveRecord: 0.0ms | Allocations: 548234)

Unit tests fail with a confusing and unhelpful error when it is run with an invalid API key. A new unit test should be added to verify that the provided API key validates before proceeding to other unit tests.

relates to #14